Bio
I am a PhD student at Gatsby Computational Neuroscience Unit UCL, advised by Peter Latham and Andrew Saxe. I use theoretical approaches to study how neural networks with different architectures [1] learn, including attention-based [2], fully-connected [3], and multimodal [4] networks.
Paper
| [1] |
 |
Saddle-to-Saddle Dynamics Explains A Simplicity Bias Across Neural Network Architectures
Yedi Zhang, Andrew Saxe, Peter E. Latham
Preprint 2025
webpage | arxiv
|
| [2] |
 |
Training Dynamics of In-Context Learning in Linear Attention
Yedi Zhang, Aaditya K. Singh, Peter E. Latham*, Andrew Saxe*
ICML 2025 (Spotlight)
pmlr | openreview | arxiv | code |
talk
|
| [3] |
 |
When Are Bias-Free ReLU Networks Effectively Linear Networks?
Yedi Zhang, Andrew Saxe, Peter E. Latham
TMLR 2025
tmlr | arxiv
|
| [4] |
 |
Understanding Unimodal Bias in Multimodal Deep Linear Networks
Yedi Zhang, Peter E. Latham, Andrew Saxe
ICML 2024
webpage | pmlr | arxiv | code
|
Blog
2025-10-10
Exponential Family (teaching notes)
2024-09-17
Eigenvalue Perturbation Theorem
2024-04-16
Isserlis' Theorem
2024-01-21
My Cribsheet for Dynamical Systems
2023-04-06
Free Energy and EM Algorithm
2023-03-18
A Manual for Reading Independence
Fun
2025-12-29
Shows I Caught
2025-12-24
Theatre Ticket Deals in London
|
